PhD graduated
Team : SMA
Departure date : 02/14/2012


An Agent Oriented Programming Language integrating Temporal Planning and the Plan Coordination Mechanisms

Over the years a lot of research has been carried out on designing new languages and platforms to program intelligent and autonomous agents. As a result, now we have the necessary tools available to develop autonomous, intelligent, adaptive, communicating and mobile agents. Most of these languages do not give agents the ability to plan ahead. But, sometimes the execution of actions without planning results in the inability to achieve the goals. Moreover, the duration of agent actions and the uncertainty of the environment has not been taken into account in the planning based agent oriented programming (AOP) languages. This thesis tries to fill this gap by proposing an AOP language P-CLAIM that endows the agents with planning capability. We are interested in the temporal planning of on the fly goals having different priorities. A coherrent framework is proposed in which agents are able to generate, execute and monitor their temporal plans. A plan is repaired if some unanticipated changes in the environment cause the plan to become unfeasible. Moreover, the proposed framework creates a balance between reactivity and deliberation. Handling and the coordination of plans for the achievement of different priority goals have not been discussed in either of the multi-agent planning and AOP languages communities. So this thesis also proposes coordination mechanisms for the plans of different priorities in two different scenarios. In the first scenario, that we call Proactive-Reactive Coordination Problem (PRCP), an agent has to modify its temporal plan in order to remove any conflicts with the plan of another agent having higher priority. This thesis proposes a plan merging algorithm supported by a sound plan repairing technique to solve this problem. In the second scenario, that we call Coordinated Planning Problem (CPP), an agent has to compute a plan for the achievement of its own goals, but without violating the constraints of another agent's higher priority plan, and utilizing where possible the cooperative opportunities offered by the latter. This thesis presents two multi-agent planners to solve this planning problem. First planner Coordinated-Sapa is an extension of the well known temporal planner Sapa, and it solves CPP for the temporal domains. The second planner µ-SATPLAN is an extension of the well known classical planner SATPLAN, and it solves CPP for non-temporal classical domains. The techniques are presented for both the planners to handle the negative (conflicting situations) as well as positive interactions (cooperative situations).

Defence : 01/24/2012

Jury members :

Mme. Adina Magda FLOREA, Professeur à l'Université Politehnica de Bucarest [Rapporteur]
M. René MANDIAU, Professeur à l'Université de Valenciennes et du Hainaut-Cambrésis [Rapporteur]
M. Rachid ALAMI, Directeur de recherche CNRS au LAAS - Toulouse
M. Humbert FIORINO, Maître de conférence à l'Université Joseph Fourier - Grenoble
M. Nicolas MAUDET, Professeur à l'Université Pierre et Marie Curie

Departure date : 02/14/2012

2009-2012 Publications